Search

This paper addresses user-specific dialogs. In contrast to previous research on personalized dialogue focused on achieving virtual user dialogue as defined by persona descriptions, user-specific dialogue aims to reproduce real-user dialogue beyond persona-based dialogue. Fine-tuning using the target user's dialogue history is an efficient learning method for a user-specific model. However, it is prone to overfitting and model destruction due to the small amount of data. Therefore, we propose a learning method for user-specific models by combining parameter-efficient fine-tuning with a pre-trained dialogue model that includes user profiles. Parameter-efficient fine-tuning adds a small number of parameters to the entire model, so even small amounts of training data can be trained efficiently and are robust to model destruction. In addition, the pre-trained model, which is learned by adding simple prompts for automatically inferred user profiles, can generate speech with enhanced knowledge of the user's profile, even when there is little training data during fine-tuning. In experiments, we compared the proposed model with large-language-model utterance generation using prompts containing users' personal information. Experiments reproducing real users' utterances revealed that the proposed model can generate utterances with higher reproducibility than the compared methods, even with a small model.

While precise data observation is essential for the learning processes of predictive models, it can be challenging owing to factors such as insufficient observation accuracy, high collection costs, and privacy constraints. In this paper, we examines cases where some qualitative features are unavailable as precise information indicating "what it is," but rather as complementary information indicating "what it is not." We refer to features defined by precise information as ordinary features (OFs) and those defined by complementary information as complementary features (CFs). We then formulate a new learning scenario termed Complementary Feature Learning (CFL), where predictive models are constructed using instances consisting of OFs and CFs. The simplest formalization of CFL applies conventional supervised learning directly using the observed values of CFs. However, this approach does not resolve the ambiguity associated with CFs, making learning challenging and complicating the interpretation of the predictive model's specific predictions. Therefore, we derive an objective function from an information-theoretic perspective to estimate the OF values corresponding to CFs and to predict output labels based on these estimations. Based on this objective function, we propose a theoretically guaranteed graph-based estimation method along with its practical approximation, for estimating OF values corresponding to CFs. The results of numerical experiments conducted with real-world data demonstrate that our proposed method effectively estimates OF values corresponding to CFs and predicts output labels., Comment: 16 pages, 7 figures

In large-data applications, such as the inference process of diffusion models, it is desirable to design sampling algorithms with a high degree of parallelization. In this work, we study the adaptive complexity of sampling, which is the minimal number of sequential rounds required to achieve sampling given polynomially many queries executed in parallel at each round. For unconstrained sampling, we examine distributions that are log-smooth or log-Lipschitz and log strongly or non-strongly concave. We show that an almost linear iteration algorithm cannot return a sample with a specific exponentially small accuracy under total variation distance. For box-constrained sampling, we show that an almost linear iteration algorithm cannot return a sample with sup-polynomially small accuracy under total variation distance for log-concave distributions. Our proof relies upon novel analysis with the characterization of the output for the hardness potentials based on the chain-like structure with random partition and classical smoothing techniques.

In our previous study, we demonstrated the existence of an unidentified ferroelectric smectic phase in the low-temperature region of the ferroelectric smectic A phase, where the layer spacing decreases with decreasing temperature. In the present study, we identified the phase by taking 2D X-ray diffraction images of a magnetically oriented sample while allowing it to rotate and constructed a 3D reciprocal space with the sample rotation angle as the third axis for the whole picture of the reciprocal lattice vectors originating from the smectic structure. Consequently, circular diffraction images were obtained when the reciprocal lattice vectors were evenly distributed on the conical surface at a certain inclination angle in the reciprocal space. This result provides clear evidence that the phase in question was smectic C. The polarization properties also showed that the observed smectic C phase had spontaneous polarization in the direction parallel to the director and was identified as ferroelectric smectic C. These results provide a new type of classification for liquid crystalline phases that has been established over many years and is a significant contribution to the basic science of soft matter research., Comment: 25 pages, 5 figures, Supporting Information

A recently suggested reciprocity relation states that the current transport efficiency from the junction to the cell terminal can be determined by differentiating luminescence images with respect to the terminal voltage. The validity of this relation is shown experimentally in this paper, by comparison with simultaneously measured electrical currents and simulations. Moreover, we verify that the method is applicable under various light concentrations and applied voltages, which allows us to investigate the cell in relevant conditions. Results evidence several kind of series resistances affecting the current transport efficiencies. We show that the relative contribution of those different resistances to the loss in current collection is a function of the illumination intensity.

Semi-supervised multi-label learning (SSMLL) is a powerful framework for leveraging unlabeled data to reduce the expensive cost of collecting precise multi-label annotations. Unlike semi-supervised learning, one cannot select the most probable label as the pseudo-label in SSMLL due to multiple semantics contained in an instance. To solve this problem, the mainstream method developed an effective thresholding strategy to generate accurate pseudo-labels. Unfortunately, the method neglected the quality of model predictions and its potential impact on pseudo-labeling performance. In this paper, we propose a dual-perspective method to generate high-quality pseudo-labels. To improve the quality of model predictions, we perform dual-decoupling to boost the learning of correlative and discriminative features, while refining the generation and utilization of pseudo-labels. To obtain proper class-wise thresholds, we propose the metric-adaptive thresholding strategy to estimate the thresholds, which maximize the pseudo-label performance for a given metric on labeled data. Experiments on multiple benchmark datasets show the proposed method can achieve the state-of-the-art performance and outperform the comparative methods with a significant margin.

In this study, we investigate the decoherence of a spatially superposed electrically neutral spin-$\frac12$ particle in the presence of a relativistic quantum electromagnetic field in Minkowski spacetime. We demonstrate that decoherence due to the spin-magnetic field coupling can be categorized into two distinct factors: local decoherence, originating from the two-point correlation functions along each branch of the superposed trajectories, and nonlocal decoherence, which arises from the correlation functions between the two superposed trajectories. These effects are linked to phase damping and amplitude damping. We also show that if the quantum field is prepared in a thermal state, decoherence monotonically increases with the field temperature., Comment: (v1)17 pages, 4 figures

This paper introduces LLM-jp, a cross-organizational project for the research and development of Japanese large language models (LLMs). LLM-jp aims to develop open-source and strong Japanese LLMs, and as of this writing, more than 1,500 participants from academia and industry are working together for this purpose. This paper presents the background of the establishment of LLM-jp, summaries of its activities, and technical reports on the LLMs developed by LLM-jp. For the latest activities, visit https://llm-jp.nii.ac.jp/en/.

The three-point correlation function (3PCF) of a weak lensing shear field contains information that is complementary to that in the two-point correlation function (2PCF), which can help improve the cosmological parameters and calibrate astrophysical and observational systematics parameters. However, the application of the 3PCF to observed data has been limited due to the computational challenges of calculating theoretical predictions of the 3PCF from a bispectrum model. In this paper, we present a new method to compute the shear 3PCF efficiently and accurately. We employ the multipole expansion of the bispectrum to compute the shear 3PCF, and show that the method is substantially more efficient than direct numerical integration. We found that the multipole-based method can compute the shear 3PCF with 5% accuracy in 10 (40) seconds for the single (four) source redshift bin setup. The multipole-based method can be also used to compute the third-order aperture mass statistics quickly and accurately, accounting for the bin-averaging effect on the shear 3PCF. Our method provides a fast and robust tool for probing the underlying cosmological model with third-order statistics of weak lensing shear., Comment: 20 pages, 9 figures, 2 tables, code available at https://github.com/git-sunao/fastnc

The present study describes a facile route of zinc oxide (ZnO) grows using the solution-dependent electrostatic spray deposition (ESD) method at temperatures ranging from 300 {\deg}C - 500 {\deg}C. In this work, zinc chloride (ZnCl2) was dissolved in ethanol (CH3CH2OH) to prepare the 0.1 M concentration of 20 ml for spray solution by ESD. Adding different deionized water (H2O) ratio, three different solutions were prepared. The results reveal that adding H2O ration, suppressing the c-axis crystal growth of ZnO thin films. The adhesion of anions was believed to be responsible for this suppression. XRD texture analysis examined the preferred orientations of the (100) and (002) planes of the ZnO thin films. Microstructural parameters namely, lattice parameters, bond length, positional parameters, full width at half maximum, crystallite sizes, lattice strain, and lattice dislocation density, are investigated. This research marks a turning point in cost-effective industrial and commercial applications for ESD-deposited electronics., Comment: 20 pages, 7 figures

Modern recommender systems heavily rely on high-quality representations learned from high-dimensional sparse data. While significant efforts have been invested in designing powerful algorithms for extracting user preferences, the factors contributing to good representations have remained relatively unexplored. In this work, we shed light on an issue in the existing pair-wise learning paradigm (i.e., the embedding collapse problem), that the representations tend to span a subspace of the whole embedding space, leading to a suboptimal solution and reducing the model capacity. Specifically, optimization on observed interactions is equivalent to a low pass filter causing users/items to have the same representations and resulting in a complete collapse. While negative sampling acts as a high pass filter to alleviate the collapse by balancing the embedding spectrum, its effectiveness is only limited to certain losses, which still leads to an incomplete collapse. To tackle this issue, we propose a novel method called DirectSpec, acting as a reliable all pass filter to balance the spectrum distribution of the embeddings during training, ensuring that users/items effectively span the entire embedding space. Additionally, we provide a thorough analysis of DirectSpec from a decorrelation perspective and propose an enhanced variant, DirectSpec+, which employs self-paced gradients to optimize irrelevant samples more effectively. Moreover, we establish a close connection between DirectSpec+ and uniformity, demonstrating that contrastive learning (CL) can alleviate the collapse issue by indirectly balancing the spectrum. Finally, we implement DirectSpec and DirectSpec+ on two popular recommender models: MF and LightGCN. Our experimental results demonstrate its effectiveness and efficiency over competitive baselines.

We present cosmological constraints using the abundance of weak-lensing shear-selected galaxy clusters in the Hyper Suprime-Cam (HSC) Subaru Strategic Program. The clusters are selected on the mass maps constructed using the three-year (Y3) weak-lensing data with an area of $\approx500~$deg$^2$, resulting in a sample size of $129$ clusters with high signal-to-noise ratios $\nu$ of $\nu\geq4.7$. Owing to the deep, wide-field, and uniform imaging of the HSC survey, this is by far the largest sample of shear-selected clusters, in which the selection solely depends on gravity and is free from any assumptions about the dynamical state. Informed by the optical counterparts, the shear-selected clusters span a redshift range of $z\lesssim0.7$ with a median of $z\approx0.3$. The lensing sources are securely selected at $z\gtrsim0.7$ with a median of $z\approx1.3$, leading to nearly zero cluster member contamination. We carefully account for (1) the bias in the photometric redshift of sources, (2) the bias and scatter in the weak-lensing mass using a simulation-based calibration, and (3) the measurement uncertainty that is directly estimated on the mass maps using an injection-based method developed in a companion paper (Chen et al. submitted). In a blind analysis, the fully marginalized posteriors of the cosmological parameters are obtained as $\Omega_{\mathrm{m}} = 0.50^{+0.28}_{-0.24}$, $\sigma_8 = 0.685^{+0.161}_{-0.088}$, $\hat{S}_{8}\equiv\sigma_8\left(\Omega_{\mathrm{m}}/0.3\right)^{0.25} = 0.835^{+0.041}_{-0.044}$, and $\sigma_8\sqrt{\Omega_{\mathrm{m}}/0.3} = 0.993^{+0.084}_{-0.126}$ in a flat $\Lambda$CDM model. We compare our cosmological constraints with other studies, including those based on cluster abundances, galaxy-galaxy lensing and clustering, and Cosmic Microwave Background observed by $Planck$, and find good agreement at levels of $\lesssim2\sigma$. [abridged], Comment: 26 pages, 14 Figures, 1 Table. To be submitted to the Open Journal of Astrophysics. Main results are Figures 11 and 12. The cluster catalog and the cosmological chain are accessible at https://github.com/inonchiu/hsc_shear_selected_clusters

We present the first step towards deriving cosmological constraints through the abundances of galaxy clusters selected in a $510\,\mathrm{deg}^2$ weak-lensing aperture mass map, constructed with the Year-Three shear catalog from the Hyper Suprime-Cam Subaru Strategic Program. We adopt a conservative source galaxy selection to construct a sample of $129$ weak-lensing peaks with a signal-to-noise ratio above $4.7$. We use semi-analytical injection simulations to derive the selection function and the mass--observable relation of our sample. These results take into account complicated uncertainties associated with weak-lensing measurements, such as the non-uniform survey depth and the complex survey geometry, projection effects from uncorrelated large-scale structures, and the intrinsic alignment of source galaxies. We also propose a novel modeling framework to make parts of the mass--observable relation insensitive to assumed cosmological parameters. Such a framework not only offers a great computational advantage to cosmological studies, but can also benefit future astrophysical studies using shear-selected clusters. Our results are an important step towards utilizing these cluster samples that are constructed nearly independent of any baryonic assumptions in upcoming deep-and-wide lensing surveys from the Vera Rubin Observatory, Euclid, and the Nancy Grace Roman Space Telescope., Comment: 17 pages, 9 figures, submitted to OJA. The cluster catalog of the cosmological sample (signal-to-noise ratio above $4.7$) is accessible at https://github.com/inonchiu/hsc_shear_selected_clusters

The compelling goal of eradicating undesirable data behaviors, while preserving usual model functioning, underscores the significance of machine unlearning within the domain of large language models (LLMs). Recent research has begun to approach LLM unlearning via gradient ascent (GA) -- increasing the prediction risk for those training strings targeted to be unlearned, thereby erasing their parameterized responses. Despite their simplicity and efficiency, we suggest that GA-based methods face the propensity towards excessive unlearning, resulting in various undesirable model behaviors, such as catastrophic forgetting, that diminish their practical utility. In this paper, we suggest a set of metrics that can capture multiple facets of real-world utility and propose several controlling methods that can regulate the extent of excessive unlearning. Accordingly, we suggest a general framework to better reflect the practical efficacy of various unlearning methods -- we begin by controlling the unlearning procedures/unlearned models such that no excessive unlearning occurs and follow by the evaluation for unlearning efficacy. Our experimental analysis on established benchmarks revealed that GA-based methods are far from perfect in practice, as strong unlearning is at the high cost of hindering the model utility. We conclude that there is still a long way towards practical and effective LLM unlearning, and more efforts are required in this field.

It has been shown that the effectiveness of graph convolutional network (GCN) for recommendation is attributed to the spectral graph filtering. Most GCN-based methods consist of a graph filter or followed by a low-rank mapping optimized based on supervised training. However, we show two limitations suppressing the power of graph filtering: (1) Lack of generality. Due to the varied noise distribution, graph filters fail to denoise sparse data where noise is scattered across all frequencies, while supervised training results in worse performance on dense data where noise is concentrated in middle frequencies that can be removed by graph filters without training. (2) Lack of expressive power. We theoretically show that linear GCN (LGCN) that is effective on collaborative filtering (CF) cannot generate arbitrary embeddings, implying the possibility that optimal data representation might be unreachable. To tackle the first limitation, we show close relation between noise distribution and the sharpness of spectrum where a sharper spectral distribution is more desirable causing data noise to be separable from important features without training. Based on this observation, we propose a generalized graph normalization G^2N to adjust the sharpness of spectral distribution in order to redistribute data noise to assure that it can be removed by graph filtering without training. As for the second limitation, we propose an individualized graph filter (IGF) adapting to the different confidence levels of the user preference that interactions can reflect, which is proved to be able to generate arbitrary embeddings. By simplifying LGCN, we further propose a simplified graph filtering (SGFCF) which only requires the top-K singular values for recommendation. Finally, experimental results on four datasets with different density settings demonstrate the effectiveness and efficiency of our proposed methods., Comment: Accepted to KDD'24

Machine unlearning as an emerging research topic for data regulations, aims to adjust a trained model to approximate a retrained one that excludes a portion of training data. Previous studies showed that class-wise unlearning is successful in forgetting the knowledge of a target class, through gradient ascent on the forgetting data or fine-tuning with the remaining data. However, while these methods are useful, they are insufficient as the class label and the target concept are often considered to coincide. In this work, we decouple them by considering the label domain mismatch and investigate three problems beyond the conventional all matched forgetting, e.g., target mismatch, model mismatch, and data mismatch forgetting. We systematically analyze the new challenges in restrictively forgetting the target concept and also reveal crucial forgetting dynamics in the representation level to realize these tasks. Based on that, we propose a general framework, namely, TARget-aware Forgetting (TARF). It enables the additional tasks to actively forget the target concept while maintaining the rest part, by simultaneously conducting annealed gradient ascent on the forgetting data and selected gradient descent on the hard-to-affect remaining data. Empirically, various experiments under the newly introduced settings are conducted to demonstrate the effectiveness of our TARF.

Recent studies have increasingly recognized the value of analyzing the post-reconstruction galaxy power spectrum for investigations into redshift space distortion (RSD) effects. In this paper, we present a novel theoretical model for the post-reconstruction galaxy power spectrum, designed for RSD analyses. In particular, we emphasize the importance of accounting for discrete effects arising from the reconstruction displacement vector, which have been overlooked. We specifically calculate these discrete effects, i.e., shot noise terms, within the framework of standard perturbation theory at the 1-loop level. In addition, we adopt a formulation that accounts for infrared (IR) effects to accurately model the non-linear damping of the Baryon Acoustic Oscillation (BAO) signal. Our model comprehensively integrates key physical phenomena relevant to the post-reconstruction galaxy power spectrum, such as gravitational non-linearities, RSD effects, bias effects, reconstruction effects, reconstruction-specific shot-noise effects, and the non-linear damping of the reconstructed BAO signal, thereby making it applicable to post-reconstruction RSD analyses., Comment: 11 pages

Diffusion models (DMs) have shown remarkable capabilities in generating realistic high-quality images, audios, and videos. They benefit significantly from extensive pre-training on large-scale datasets, including web-crawled data with paired data and conditions, such as image-text and image-class pairs. Despite rigorous filtering, these pre-training datasets often inevitably contain corrupted pairs where conditions do not accurately describe the data. This paper presents the first comprehensive study on the impact of such corruption in pre-training data of DMs. We synthetically corrupt ImageNet-1K and CC3M to pre-train and evaluate over 50 conditional DMs. Our empirical findings reveal that various types of slight corruption in pre-training can significantly enhance the quality, diversity, and fidelity of the generated images across different DMs, both during pre-training and downstream adaptation stages. Theoretically, we consider a Gaussian mixture model and prove that slight corruption in the condition leads to higher entropy and a reduced 2-Wasserstein distance to the ground truth of the data distribution generated by the corruptly trained DMs. Inspired by our analysis, we propose a simple method to improve the training of DMs on practical datasets by adding condition embedding perturbations (CEP). CEP significantly improves the performance of various DMs in both pre-training and downstream tasks. We hope that our study provides new insights into understanding the data and pre-training processes of DMs., Comment: 50 pages, 33 figures, 4 tables

In federated learning (FL), the multi-step update and data heterogeneity among clients often lead to a loss landscape with sharper minima, degenerating the performance of the resulted global model. Prevalent federated approaches incorporate sharpness-aware minimization (SAM) into local training to mitigate this problem. However, the local loss landscapes may not accurately reflect the flatness of global loss landscape in heterogeneous environments; as a result, minimizing local sharpness and calculating perturbations on client data might not align the efficacy of SAM in FL with centralized training. To overcome this challenge, we propose FedLESAM, a novel algorithm that locally estimates the direction of global perturbation on client side as the difference between global models received in the previous active and current rounds. Besides the improved quality, FedLESAM also speed up federated SAM-based approaches since it only performs once backpropagation in each iteration. Theoretically, we prove a slightly tighter bound than its original FedSAM by ensuring consistent perturbation. Empirically, we conduct comprehensive experiments on four federated benchmark datasets under three partition strategies to demonstrate the superior performance and efficiency of FedLESAM.

Various approaches have emerged for multi-armed bandits in distributed systems. The multiplayer dueling bandit problem, common in scenarios with only preference-based information like human feedback, introduces challenges related to controlling collaborative exploration of non-informative arm pairs, but has received little attention. To fill this gap, we demonstrate that the direct use of a Follow Your Leader black-box approach matches the lower bound for this setting when utilizing known dueling bandit algorithms as a foundation. Additionally, we analyze a message-passing fully distributed approach with a novel Condorcet-winner recommendation protocol, resulting in expedited exploration in many cases. Our experimental comparisons reveal that our multiplayer algorithms surpass single-player benchmark algorithms, underscoring their efficacy in addressing the nuanced challenges of the multiplayer dueling bandit setting.

Linear Mode Connectivity (LMC) refers to the phenomenon that performance remains consistent for linearly interpolated models in the parameter space. For independently optimized model pairs from different random initializations, achieving LMC is considered crucial for validating the stable success of the non-convex optimization in modern machine learning models and for facilitating practical parameter-based operations such as model merging. While LMC has been achieved for neural networks by considering the permutation invariance of neurons in each hidden layer, its attainment for other models remains an open question. In this paper, we first achieve LMC for soft tree ensembles, which are tree-based differentiable models extensively used in practice. We show the necessity of incorporating two invariances: subtree flip invariance and splitting order invariance, which do not exist in neural networks but are inherent to tree architectures, in addition to permutation invariance of trees. Moreover, we demonstrate that it is even possible to exclude such additional invariances while keeping LMC by designing decision list-based tree architectures, where such invariances do not exist by definition. Our findings indicate the significance of accounting for architecture-specific invariances in achieving LMC.

Current Reinforcement Learning (RL) is often limited by the large amount of data needed to learn a successful policy. Offline RL aims to solve this issue by using transitions collected by a different behavior policy. We address a novel Offline RL problem setting in which, while collecting the dataset, the transition and reward functions gradually change between episodes but stay constant within each episode. We propose a method based on Contrastive Predictive Coding that identifies this non-stationarity in the offline dataset, accounts for it when training a policy, and predicts it during evaluation. We analyze our proposed method and show that it performs well in simple continuous control tasks and challenging, high-dimensional locomotion tasks. We show that our method often achieves the oracle performance and performs better than baselines., Comment: Accepted for Reinforcement Learning Conference (RLC) 2024

In mobile and IoT systems, Federated Learning (FL) is increasingly important for effectively using data while maintaining user privacy. One key challenge in FL is managing statistical heterogeneity, such as non-i.i.d. data, arising from numerous clients and diverse data sources. This requires strategic cooperation, often with clients having similar characteristics. However, we are interested in a fundamental question: does achieving optimal cooperation necessarily entail cooperating with the most similar clients? Typically, significant model performance improvements are often realized not by partnering with the most similar models, but through leveraging complementary data. Our theoretical and empirical analyses suggest that optimal cooperation is achieved by enhancing complementarity in feature distribution while restricting the disparity in the correlation between features and targets. Accordingly, we introduce a novel framework, \texttt{FedSaC}, which balances similarity and complementarity in FL cooperation. Our framework aims to approximate an optimal cooperation network for each client by optimizing a weighted sum of model similarity and feature complementarity. The strength of \texttt{FedSaC} lies in its adaptability to various levels of data heterogeneity and multimodal scenarios. Our comprehensive unimodal and multimodal experiments demonstrate that \texttt{FedSaC} markedly surpasses other state-of-the-art FL methods.

The Simons Observatory (SO) is a cosmic microwave background (CMB) survey experiment that includes small-aperture telescopes (SATs) observing from an altitude of 5,200 m in the Atacama Desert in Chile. The SO SATs will cover six spectral bands between 27 and 280 GHz to search for primordial B-modes to a sensitivity of $\sigma(r)=0.002$, with quantified systematic errors well below this value. Each SAT is a self-contained cryogenic telescope with a 35$^\circ$ field of view, 42 cm diameter optical aperture, 40 K half-wave plate, 1 K refractive optics, and $<0.1$ K focal plane that holds $>12,000$ TES detectors. We describe the nominal design of the SATs and present details about the integration and testing for one operating at 93 and 145 GHz.

In this paper, we investigate an offline reinforcement learning (RL) problem where datasets are collected from two domains. In this scenario, having datasets with domain labels facilitates efficient policy training. However, in practice, the task of assigning domain labels can be resource-intensive or infeasible at a large scale, leading to a prevalence of domain-unlabeled data. To formalize this challenge, we introduce a novel offline RL problem setting named Positive-Unlabeled Offline RL (PUORL), which incorporates domain-unlabeled data. To address PUORL, we develop an offline RL algorithm utilizing positive-unlabeled learning to predict the domain labels of domain-unlabeled data, enabling the integration of this data into policy training. Our experiments show the effectiveness of our method in accurately identifying domains and learning policies that outperform baselines in the PUORL setting, highlighting its capability to leverage domain-unlabeled data effectively.

The key to multi-label image classification (MLC) is to improve model performance by leveraging label correlations. Unfortunately, it has been shown that overemphasizing co-occurrence relationships can cause the overfitting issue of the model, ultimately leading to performance degradation. In this paper, we provide a causal inference framework to show that the correlative features caused by the target object and its co-occurring objects can be regarded as a mediator, which has both positive and negative impacts on model predictions. On the positive side, the mediator enhances the recognition performance of the model by capturing co-occurrence relationships; on the negative side, it has the harmful causal effect that causes the model to make an incorrect prediction for the target object, even when only co-occurring objects are present in an image. To address this problem, we propose a counterfactual reasoning method to measure the total direct effect, achieved by enhancing the direct effect caused only by the target object. Due to the unknown location of the target object, we propose patching-based training and inference to accomplish this goal, which divides an image into multiple patches and identifies the pivot patch that contains the target object. Experimental results on multiple benchmark datasets with diverse configurations validate that the proposed method can achieve state-of-the-art performance.

Polynomials commute under composition are referred to as commuting polynomials. In this paper, we study division properties for commuting polynomials with rational (and integer) coefficients. As a consequence, we show an algebraic particularity of the commuting polynomials coming from weighted sums for cycle graphs with pendant edges (arXiv:2402.07209v1.)., Comment: 16 pages

The baryonic feedback effect is considered as a possible solution to the so-called $S_8$ tension indicated in cosmic shear cosmology. The baryonic effect is more significant on smaller scales, and affects the cosmic shear two-point correlation functions (2PCFs) with different scale- and redshift-dependencies from those of the cosmological parameters. In this paper, we use the Hyper Suprime-Cam Year 3 (HSC-Y3) data to measure the cosmic shear 2PCFs ($\xi_{\pm}$) down to 0.28 arcminutes, taking full advantage of the high number density of source galaxies in the deep HSC data, to explore a possible signature of the baryonic effect. While the published HSC analysis used the cosmic shear 2PCFs on angular scales, which are sensitive to the matter power spectrum at $k\lesssim 1~h{\rm Mpc}^{-1}$, the smaller scale HSC cosmic shear signal allows us to probe the signature of matter power spectrum up to $k\simeq 20~h{\rm Mpc}^{-1}$. Using the accurate emulator of the nonlinear matter power spectrum, DarkEmulator2, we show that the dark matter-only model can provide an acceptable fit to the HSC-Y3 2PCFs down to the smallest scales. In other words, we do not find any clear signature of the baryonic effects or do not find a systematic shift in the $S_8$ value with the inclusion of the smaller-scale information as would be expected if the baryonic effect is significant. Alternatively, we use a flexible 6-parameter model of the baryonic effects, which can lead to both enhancement and suppression in the matter power spectrum compared to the dark matter-only model, to perform the parameter inference of the HSC-Y3 2PCFs. We find that the small-scale HSC data allow only a fractional suppression of up to 5 percent in the matter power spectrum at $k\sim 1~h{\rm Mpc}^{-1}$, which is not sufficient to reconcile the $S_8$ tension., Comment: 30 pages, 16 figures

While substantial progress has been made in studying the pre-reconstruction galaxy bispectrum, investigations of the post-reconstruction bispectrum are still in nascent stages. In this paper, we present a bispectrum model that incorporates one-loop corrections in the Standard Perturbation Theory (SPT), while simultaneously addressing infrared (IR) effects in the post-reconstruction density fluctuations in a non-perturbative approach. This model accurately captures the nonlinear behavior of the Baryon Acoustic Oscillation (BAO) signal within the post-reconstruction bispectrum framework. Furthermore, the incorporation of the one-loop correction extends its applicability to smaller scales. Given that the approach to addressing IR effects remains invariant before and after the reconstruction process, the resulting IR-resummed bispectrum model exhibits uniformity across both scenarios. Throughout this analysis, we achieve a comprehensive and unified description of the bispectrum that seamlessly spans the pre- and post-reconstruction phases. An explanatory video is available at https://youtu.be/bHdlIi3jQ4A., Comment: 15 pages. An explanatory video is available at https://youtu.be/bHdlIi3jQ4A

The current thesis aims to explore the reinforcement learning field and build on existing methods to produce improved ones to tackle the problem of learning in high-dimensional and complex environments. It addresses such goals by decomposing learning tasks in a hierarchical fashion known as Hierarchical Reinforcement Learning. We start in the first chapter by getting familiar with the Markov Decision Process framework and presenting some of its recent techniques that the following chapters use. We then proceed to build our Hierarchical Policy learning as an answer to the limitations of a single primitive policy. The hierarchy is composed of a manager agent at the top and employee agents at the lower level. In the last chapter, which is the core of this thesis, we attempt to learn lower-level elements of the hierarchy independently of the manager level in what is known as the "Eigenoption". Based on the graph structure of the environment, Eigenoptions allow us to build agents that are aware of the geometric and dynamic properties of the environment. Their decision-making has a special property: it is invariant to symmetric transformations of the environment, allowing as a consequence to greatly reduce the complexity of the learning task., Comment: Master Thesis 2018, MVA ENS Paris-Saclay, Tokyo RIKEN AIP

Foundation models are usually pre-trained on large-scale datasets and then adapted to downstream tasks through tuning. However, the large-scale pre-training datasets, often inaccessible or too expensive to handle, can contain label noise that may adversely affect the generalization of the model and pose unexpected risks. This paper stands out as the first work to comprehensively understand and analyze the nature of noise in pre-training datasets and then effectively mitigate its impacts on downstream tasks. Specifically, through extensive experiments of fully-supervised and image-text contrastive pre-training on synthetic noisy ImageNet-1K, YFCC15M, and CC12M datasets, we demonstrate that, while slight noise in pre-training can benefit in-domain (ID) performance, where the training and testing data share a similar distribution, it always deteriorates out-of-domain (OOD) performance, where training and testing distributions are significantly different. These observations are agnostic to scales of pre-training datasets, pre-training noise types, model architectures, pre-training objectives, downstream tuning methods, and downstream applications. We empirically ascertain that the reason behind this is that the pre-training noise shapes the feature space differently. We then propose a tuning method (NMTune) to affine the feature space to mitigate the malignant effect of noise and improve generalization, which is applicable in both parameter-efficient and black-box tuning manners. We additionally conduct extensive experiments on popular vision and language models, including APIs, which are supervised and self-supervised pre-trained on realistic noisy data for evaluation. Our analysis and results demonstrate the importance of this novel and fundamental research direction, which we term as Noisy Model Learning., Comment: 18 pages, 10 figures, 6 tables, preprint. arXiv admin note: substantial text overlap with arXiv:2309.17002

Data augmentation is an area of research which has seen active development in many machine learning fields, such as in image-based learning models, reinforcement learning for self driving vehicles, and general noise injection for point cloud data. However, convincing methods for general time series data augmentation still leaves much to be desired, especially since the methods developed for these models do not readily cross-over. Three common approaches for time series data augmentation include: (i) Constructing a physics-based model and then imbuing uncertainty over the coefficient space (for example), (ii) Adding noise to the observed data set(s), and, (iii) Having access to ample amounts of time series data sets from which a robust generative neural network model can be trained. However, for many practical problems that work with time series data in the industry: (i) One usually does not have access to a robust physical model, (ii) The addition of noise can in of itself require large or difficult assumptions (for example, what probability distribution should be used? Or, how large should the noise variance be?), and, (iii) In practice, it can be difficult to source a large representative time series data base with which to train the neural network model for the underlying problem. In this paper, we propose a methodology which attempts to simultaneously tackle all three of these previous limitations to a large extent. The method relies upon the well-studied matrix differential geometry of the Stiefel manifold, as it proposes a simple way in which time series signals can placed on, and then smoothly perturbed over the manifold. We attempt to clarify how this method works by showcasing several potential use cases which in particular work to take advantage of the unique properties of this underlying manifold., Comment: 61 pages, 41 figures

The celebrated result of Gortler-Healy-Thurston (independently, Jackson-Jord\'an for $d=2$) shows that the global rigidity of graphs realised in the $d$-dimensional Euclidean space is a generic property. Extending this result to the global rigidity problem in $\ell_p$-spaces remains an open problem. In this paper we affirmatively solve this problem when $d=2$ and $p$ is an even positive integer. A key tool in our proof is a sufficient condition for the $d$-tangentially weakly non-defectiveness of projective varieties due to Bocci, Chiantini, Ottaviani, and Vannieuwenhoven. By specialising the condition to the $p$-Cayley-Menger variety, which is the $\ell_p$-analogue of the Cayley-Menger variety for Euclidean distance, we provide an $\ell_p$-extension of the generic global rigidity theory of Connelly. As a by-product of our proof, we also offer a purely graph-theoretical characterisation of the $2$-identifiability of an orthogonal projection of the $p$-Cayley-Menger variety along a coordinate axis of the ambient affine space.

Social networks are often associated with rich side information, such as texts and images. While numerous methods have been developed to identify communities from pairwise interactions, they usually ignore such side information. In this work, we study an extension of the Stochastic Block Model (SBM), a widely used statistical framework for community detection, that integrates vectorial edges covariates: the Vectorial Edges Covariates Stochastic Block Model (VEC-SBM). We propose a novel algorithm based on iterative refinement techniques and show that it optimally recovers the latent communities under the VEC-SBM. Furthermore, we rigorously assess the added value of leveraging edge's side information in the community detection process. We complement our theoretical results with numerical experiments on synthetic and semi-synthetic data.

The Simons Observatory (SO) experiment is a cosmic microwave background (CMB) experiment located in the Atacama Desert, Chile. The SO' s small aperture telescopes (SATs) consist of three telescopes designed for precise CMB polarimetry at large angular scales. Each SAT uses a cryogenic rotating half-wave plate (HWP) as a polarization modulator to mitigate atmospheric 1/f noise and other systematics. To realize efficient polarization modulation over the observation bands, we fabricated an achromatic HWP (AHWP) consisting of three sapphire plates with anti-reflection coatings. The AHWP is designed to have broadband modulation efficiency and transmittance. This paper reports on the design and the preliminary characterization of the AHWPs for SATs.

To improve the ability of the large language model (LLMs) to tackle complex reasoning problems, chain-of-thoughts (CoT) methods were proposed to guide LLMs to reason step-by-step, enabling problem solving from simple to complex. State-of-the-art methods for generating such a chain involve interactive collaboration, where the learner generates candidate intermediate thoughts, evaluated by the LLM, guiding the generation of subsequent thoughts. However, a widespread yet understudied problem is that the evaluation from the LLM is typically noisy and unreliable, potentially misleading the generation process in selecting promising intermediate thoughts. In this paper, motivated by Vapnik's principle, we use pairwise-comparison evaluation instead of point-wise scoring to search for promising intermediate thoughts with the noisy feedback from the LLM. In each round, we randomly pair intermediate thoughts and directly prompt the LLM to select the more promising one from each pair, allowing us to identify the most promising thoughts through an iterative process. To further alleviate the noise in the comparison, we incorporate techniques from ensemble learning and dueling bandits, proposing two variants of the algorithm. Experiments on three real-world tasks demonstrate the effectiveness of our proposed algorithm and verify the rationale of the pairwise comparison mechanism., Comment: ICML 2024

Since galaxy distribution reconstruction effectively reduces non-Gaussian terms in the power spectrum covariance matrix, it has attracted interest not only for Baryon Acoustic Oscillation (BAO) signals but also for various cosmological signal analyses. To this end, this paper presents a novel theoretical model that addresses infrared (IR) effects in the post-reconstruction galaxy power spectrum, including 1-loop corrections. In particular, we discuss the importance of incorporating non-perturbative effects arising from IR contributions into the displacement vector $\vec{s}$ used for reconstruction. Consequently, post-reconstruction nonlinear damping of BAO can be described by a single two-dimensional Gaussian function. This is a phenomenon not observed when $\vec{s}$ is considered to at a linear order in the Zel'dovich approximation. Furthermore, we confirm that the cross-power spectrum of the pre- and post-reconstruction density fluctuations lacks IR effect cancellations, and shows an exponential decay in both the cross-power spectrum and the associated shot-noise term. An explanatory video is available at https://youtu.be/u1-xx3_4xCg, Comment: 19 pages. An explanatory video is available at https://youtu.be/u1-xx3_4xCg

In Reinforcement Learning (RL), it is commonly assumed that an immediate reward signal is generated for each action taken by the agent, helping the agent maximize cumulative rewards to obtain the optimal policy. However, in many real-world scenarios, immediate reward signals are not obtainable; instead, agents receive a single reward that is contingent upon a partial sequence or a complete trajectory. In this work, we define this challenging problem as Reinforcement Learning from Bagged Reward (RLBR), where sequences of data are treated as bags with non-Markovian bagged rewards. We provide a theoretical study to establish the connection between RLBR and standard RL in Markov Decision Processes (MDPs). To effectively explore the reward distributions within these bags and enhance policy training, we propose a Transformer-based reward model, the Reward Bag Transformer, which employs a bidirectional attention mechanism to interpret contextual nuances and temporal dependencies within each bag. Our empirical evaluations reveal that the challenge intensifies as the bag length increases, leading to the performance degradation due to reduced informational granularity. Nevertheless, our approach consistently outperforms existing methods, demonstrating the least decline in efficacy across varying bag lengths and excelling in approximating the original MDP's reward distribution.

Weakly supervised learning generally faces challenges in applicability to various scenarios with diverse weak supervision and in scalability due to the complexity of existing algorithms, thereby hindering the practical deployment. This paper introduces a general framework for learning from weak supervision (GLWS) with a novel algorithm. Central to GLWS is an Expectation-Maximization (EM) formulation, adeptly accommodating various weak supervision sources, including instance partial labels, aggregate statistics, pairwise observations, and unlabeled data. We further present an advanced algorithm that significantly simplifies the EM computational demands using a Non-deterministic Finite Automaton (NFA) along with a forward-backward algorithm, which effectively reduces time complexity from quadratic or factorial often required in existing solutions to linear scale. The problem of learning from arbitrary weak supervision is therefore converted to the NFA modeling of them. GLWS not only enhances the scalability of machine learning models but also demonstrates superior performance and versatility across 11 weak supervision scenarios. We hope our work paves the way for further advancements and practical deployment in this field., Comment: 24 pages, 20 tables, 9 figures

We prove that Hecke eigenvalues for any Hilbert and Siegel modular forms are algebraic integers. Our method does not rely on cohomologicality nor Galois representations. We apply the integrality of Hecke eigenvalues for Hilbert modular forms of non-parallel weight to the estimation of the growth of Hecke fields of Hilbert cusp forms with non-vanishing central $L$-values. As a further application, we give the growth of the fields of rationality of cuspidal automorphic representations of ${\rm GL}_{2d}(\mathbb{A}_\mathbb{Q})$ for a prime number $d$ with non-vanishing central $L$-values. We also apply the integrality of Hecke eigenvalues for holomorphic Siegel cusp forms of general degree in order to give the growth of the Hecke fields of those forms.

The Extreme Universe Space Observatory on a Super Pressure Balloon 1 (EUSO-SPB1) was launched in 2017 April from Wanaka, New Zealand. The plan of this mission of opportunity on a NASA super pressure balloon test flight was to circle the southern hemisphere. The primary scientific goal was to make the first observations of ultra-high-energy cosmic-ray extensive air showers (EASs) by looking down on the atmosphere with an ultraviolet (UV) fluorescence telescope from suborbital altitude (33~km). After 12~days and 4~hours aloft, the flight was terminated prematurely in the Pacific Ocean. Before the flight, the instrument was tested extensively in the West Desert of Utah, USA, with UV point sources and lasers. The test results indicated that the instrument had sensitivity to EASs of approximately 3 EeV. Simulations of the telescope system, telescope on time, and realized flight trajectory predicted an observation of about 1 event assuming clear sky conditions. The effects of high clouds were estimated to reduce this value by approximately a factor of 2. A manual search and a machine-learning-based search did not find any EAS signals in these data. Here we review the EUSO-SPB1 instrument and flight and the EAS search., Comment: 18 pages, 19 figures

Understanding the phenomenon of quantum superposition of gravitational fields induced by massive quantum particles is an important starting point for quantum gravity. The purpose of this study is to deepen our understanding of the phenomenon of quantum superposition of gravitational fields. To this end, we consider a trade-off relation of entanglement (monogamy relation) in a tripartite system consisting of two massive particles and a gravitational field that may be entangled with each other. Consequently, if two particles cannot exchange information mutually, they are in a separable state, and the particle and gravitational field are always entangled. Furthermore, even when two particles can send information to each other, there is a trade-off between the two particles and the gravitational field. We also investigate the behavior of the quantum superposition of the gravitational field using quantum discord. We find that quantum discord increases depending on the length scale of the particle superposition. Our results may help understand the relationship between the quantization of the gravitational field and the meaning of the quantum superposition of the gravitational field., Comment: 19 pages, 6 figures